问题原文
原帖链接 · JKnight · CC BY-SA 3.0
My understanding of RAID 0 is that the data is split and striped across all of the drives in the array. If there is only one drive in the array, then what is the point of a single drive RAID 0 array and how does it work?
From these ceph benchmarks it appears there is a tangible performance difference as compared to JBOD in at least some cases.
Despite my best Googling I could not find anything substantial. Thanks!
已采纳解答
原帖链接 · Anton Cohen · CC BY-SA 3.0
In the case of this implementation (the LSI SAS2208 controller), JBOD does not use on-board cache, single disk RAID0 uses on-board write-back cache. The ceph benchmark explains it in the test setup. The performance increase comes from caching not striping. Most RAID controllers allow you to setup single disk RAID0 or RAID1 as a way to support JBOD, this controller is a little different in that it also supports JBOD (without any controller cache).
There are cases where RAID levels can use less disks than you normally think, and still provide increased performance or redundancy. For example Linux md RAID10 can be used with two or more disks, including odd numbers of disks, unlike traditional RAID10 which would require four or more disks and even numbers. Linux md RAID10 with two disks is faster than RAID1.
摘自公开归档,保留原文语言,版式经整理,外部媒体与邮箱地址已移除。
